COIOS
Weighing data and models Item confirms peer-reviewed

A random forest beat logistic regression at predicting death in US diabetic adults

In 2428 US adults with both type 2 diabetes and hypertension from NHANES 1999-2018, a random forest model predicted all-cause mortality with an AUC of 0.873, against 0.783 for logistic regression, with no external validation reported.

The authors used NHANES 1999 to 2018 linked to the National Death Index to build five models predicting all-cause mortality in 2,428 adults with both type 2 diabetes and hypertension, among whom 719 (29.6%) died over a median 6.75 years. Random forest reached an AUC of 0.873 (95% CI 0.856 to 0.891), above light gradient boosting (0.785), XGBoost (0.792) and logistic regression (0.783), with age, race, chronic kidney disease, BMI and blood urea nitrogen the leading features. Validation was internal to the survey sample.

Why it is interesting: A tree ensemble outperformed logistic regression on this cohort, though no external cohort was tested to confirm it.

Source
JMIR Medical Informatics, 22 September 2026
DOI
10.2196/85557
Type
Peer-reviewed article
Design
Prediction model development on NHANES 1999-2018 linked to deaths to 2019; n=2428 adults with type 2 diabetes and hypertension, 719 deaths, median follow-up 6.75 years
Verdict
Confirms prior evidence
Driver
AI in medicine
Driver
Obesity, diabetes and GLP-1
Driver
Health funding and inequalities